Internet memes are a powerful form of online communication, yet their multimodal nature and reliance on commonsense and cultural knowledge make automated toxicity detection challenging. Prior work has studied individual dimensions of memes: textual content via optical character recognition, visual content via object recognition, and toxicity via hate speech detection, but lacks a unified architecture that formally identifies the semantic dimensions contributing to meme toxicity. We present three contributions: (1) a semantic framework identifying the dimensions necessary for meme interpretation; (2) a corresponding benchmark, M-QUEST, of 609 question-answer pairs for 307 memes, generated via a semi-automatic process grounded in the framework; and (3) an empirical evaluation of 12 VLMs on M-QUEST. The results show that the instruction-tuned models with reasoning capabilities significantly outperform others, whereas pragmatic inference remains challenging across all models.

Fact-checking claims is a highly laborious task that involves understanding how each factual assertion within the claim relates to a set of trusted source materials. Existing approaches make sample-level predictions but fail to identify the specific aspects of the claim that are troublesome and the specific evidence relied upon. In this paper, we introduce a method and new benchmark for this challenging task. Our method predicts the fine-grained logical relationship of each aspect of the claim from a set of multimodal documents, which include text, image(s), video(s), and audio(s). We also introduce a new benchmark (M3DC) of claims requiring multimodal multidocument reasoning, which we construct using a novel claim synthesis technique. Experiments show that our approach outperforms other models on this challenging task on two benchmarks while providing finer-grained predictions, explanations, and evidence.

Fact-checking real-world claims often requires reviewing multiple multimodal documents in order to assess the claim’s truthfulness, a highly laborious and time-consuming task. In this paper, we present a summarization model crafted to generate claim-specific summaries useful for fact-checking from multimodal multi-document datasets. The model takes inputs in the form of documents, images, and a claim, with the objective of assisting in fact-checking tasks. We introduce a dynamic perceiver-based model that is able to handle inputs from multiple modalities of arbitrary lengths. To train our model, we leverage a novel reinforcement learning-based entailment objective in order to generate summaries that provide evidence distinguishing between different truthfulness labels. To assess the efficacy of our approach, we conduct experiments on both an existing benchmark as well as a new dataset of multi-document claims which we contribute. Our approach outperforms the SOTA approach by 4.6% in the claim verification task on the MOCHEG dataset and demonstrates strong performance on our new Multi-News-Fact-Checking dataset.